Express Sum and Carry as a function of P, G, DDefine 3 new variable which ONLY depend on A, B... The Ripple-Carry AdderWorst case delay linear with the number of bits Goal: Make the fast
Trang 1Digital Integrated Circuits
A Design Perspective
Arithmetic Circuits
Jan M Rabaey Anantha Chandrakasan Borivoje Nikolic
January, 2003
Trang 2A Generic Digital Processor
Trang 3Building Blocks for Digital Architectures
Trang 4b
s0 s1
g64
LU : Logical Unit
Trang 5Bit-Sliced Design
Bit 3Bit 2Bit 1Bit 0
Trang 6Bit-Sliced Datapath
Adder stage 1 Wiring Adder stage 2
Trang 7Itanium Integer Datapath
Trang 8Adders
Trang 9Cout Sum
adder
Trang 10The Binary Adder
Cin Full
adder
Trang 11Express Sum and Carry as a function of P, G, D
Define 3 new variable which ONLY depend on A, B
Trang 12The Ripple-Carry Adder
Worst case delay linear with the number of bits
Goal: Make the fastest possible carry path circuit
Trang 13Complimentary Static CMOS Full Adder
28 Transistors
B A
C i
C i A B
C o
V DD
Trang 15Minimize Critical Path by Reducing Inverting Stages
Exploit Inversion Property
Trang 16A Better Structure: The Mirror Adder
V DD
C i
A B B
A
B A
C i
A B B
Trang 18The Mirror Adder
A maximum of two series transistors can be observed in the generation circuitry.
carry-•When laying out the cell, the most critical issue is the
minimization of the capacitance at node Co The reduction of the diffusion capacitances is particularly important.
capacitances, two internal gate capacitances, and six gate
capacitances in the connecting adder cell
•Only the transistors in the carry stage have to be optimized for
optimal speed All transistors in the sum stage can be minimal
Trang 19Transmission Gate Full Adder
A
B P
P
Sum Generation
Carry Generation
Setup
Trang 20Manchester Carry Chain
Trang 21Manchester Carry Chain
Trang 22Manchester Carry Chain
Trang 24Carry-Bypass Adder (cont.)
Carry propagation
Setup Bit 0–3
Setup Bit 4–7
Sum
t bypass
Carry propagation
Setup Bit 8–11
Sum
Carry propagation
Setup Bit 12–15
Sum
Trang 25Carry Ripple versus Carry Bypass
Trang 27Carry Select Adder: Critical Path
0
1
Sum Generation
Multiplexer 1-Carry 0-Carry Setup
0
1
Sum Generation Multiplexer 1-Carry
S
Trang 28Linear Carry Select
Trang 29Square Root Carry Select
Trang 30Adder Delays - Comparison
Square root select
Linear selectRipple adder
100
20304050
Trang 31LookAhead - Basic Idea
Trang 32Look-Ahead: Topology
Co k, = Gk+ Pk( Gk 1– + Pk 1– Co k 2, – )
Expanding Lookahead equations:
All the way:
Trang 33Logarithmic Look-Ahead Adder
Trang 34Carry Lookahead Trees
Trang 39Example: Domino Adder
Trang 40Example: Domino Adder
Trang 41Example: Domino Sum
Trang 42Multipliers
Trang 43The Binary Multiplication
Trang 44The Binary Multiplication
x
+
Partial products
Multiplicand Multiplier
Trang 45The Array Multiplier
Trang 46The MxN Array Multiplier
— Critical Path
HA FA FA HA
HA FA
FA FA
FA
Critical Path 1 Critical Path 2
Critical Path 1 & 2
Trang 47Carry-Save Multiplier
FA FA
FA HA
Trang 48Multiplier Floorplan
S C S
C S
C S
C
S C S
C S
C S
C
S C S
C S
C S
C
S
C S
C S
C S
Trang 52Multipliers —Summary
• Optimization Goals Different Vs Binary Adder
• Once Again: Identify Critical Path
• Other possible techniques
- Data encoding (Booth)
- Pipelining FIRST GLIMPSE AT SYSTEM LEVEL OPTIMIZATION
- Logarithmic versus Linear (Wallace Tree Mult)
Trang 53Shifters
Trang 54The Binary Shifter
Trang 55The Barrel Shifter
Sh3 Sh2
Sh1 Sh0
Sh3 Sh2
Area Dominated by Wiring
Trang 564x4 barrel shifter
Buffer
Sh3 Sh2
Sh1 Sh0